The potential impact of emotionally loaded stimuli on over/under-estimating neutral situations among power plant control-room operators

1 Introduction

In the last few decades, the science of emotions has dramatically evolved [1]. Moreover, the study of emotions can be traced back to Darwin’s ideas about how emotions are natural or shared by all people [2]. However, as a result of rapid advancement in technology and more interactions between humans and computers, the focus of research communities has shifted significantly towards emotions and issues related to emotions [3]. It is argued that humans have an innate core set of emotions that are universally expressed and recognized [4]. Humans use emotions to unravel intricate cascades of internal events [5].

Since the situations in which individuals communicate their emotions differ greatly from one another, there is a significant amount of complexity involved in studying and recognizing emotional responses [6].

Emotions that are defined as inner feelings, such as the experience of the state of fear, anger, and shame [7], can affect negotiation, confidence, reactions to ultimatums, collaboration in public goods dilemmas, conflict resolution in interdependent decision-making, and mixed-motive interactions where outcomes are dependent on their own and others’ decisions [8]. Moreover, emotions do not inherently constitute a mode of heuristic thinking. Initially, emotions are easily elicited and can cause swift action [1]. Another research provides insight into the interaction of activities and emotions in forming human behavior, leading to significant practical consequences [9].

Emotional experiences appear to be important in psychosocial development and mental health across the lifespan [2]. As emotions have causal effects on the nature of our relationships, sleep habits, decisions, political and policy choices, innovations, physical and mental health, health and safety in the workplace and the general wellbeing, the subject deserves to be continuously scrutinized [1, 10].

Additionally, the field of emotion research has attracted much attention since the quantification of emotional states, such as joy, enthusiasm, and anxiety, expands our understanding of such qualia between humans and computers [11]. With this knowledge, the chance of improving both machine and human training (as two parts of human-computer interaction) increases, so errors are reduced and performance is improved.

Despite high automation level, humans always play a crucial role in process industries since errors are a possible outcome of complex human response. An incorrect diagnosis can lead to the development of new failure mechanisms most of which have not been considered in ordinary risk assessments [12].

The emotional job requirements can be described as a dependent component of emotional labor. There is evidence that emotional labor along with improper managing and controlling of emotions– having less information about the emotions, and difficulty in knowing, naming, and labeling the emotions can be considered parts of this process- is associated positively with strain and negatively with performance in the workplace [13].

Emotions control behavior (i.e., approach or avoidance), affect numerous cognitive processes (including memory and decision making, evaluations, etc.) and signify the presence of important events in the environment [14, 15]. Based on one of the most attractive aspects of related science about how humans deal with limited resources in a changing world, the human brain is severely capacity constrained, despite its amazing complexity and processing power [20].

It should be noted that all of the data that enters the sensory system cannot be adequately processed and represented [20], so some of the data may be ignored or magnified based on the context.

Meanwhile, emotion has a considerable impact on individuals’ performance in their activities of daily living [16]. There is a piece of growing evidence that not only decision-making can be affected by emotional states [9], but there seem to be robust contextual effects of emotion on individuals’ decision-making [8]. Since emotions are strong, pervasive, predictable, often harmful, and sometimes beneficial drivers of decision-making [1], the nature of emotional interactions might be considered as an essential parameter in the workplace, especially where human-machine interaction is a crucial indicator of productivity [17].

Performance is a critical issue in all industries, especially in process industries, such as power plants and their control rooms as the beating heart of the industry, where an error may cause irrecoverable damage and claim lives leading to large financial losses. In addition to routine situations in the workplace and control rooms, abnormal situations such as emergencies are matters of concern, especially because human intervention is essential and inevitable. Moreover, these situations have to be taken into consideration because of complexities in both human and system. The same is true with start-up, shut-down, grade changes, etc. Consequently, the critical role of emotion and its effect on performance can be considered an inevitable parameter in industries.

Based on recent studies, the identification of emotions can be used to favorably predict successful outcomes in various aspects of an organization. Furthermore, it can be confirmed that the ability to identify emotions is an essential skill in any profession [18].

There may also be problems and difficulties in the workplace related to emotions and emotion identification, for example, when employees cannot understand the emotional state or when they overestimate the emotions. Evidence from various areas of workplace emotion research indicates the importance of distinguishing between different kinds of emotions when studying correlations between emotional tendencies and performance [18].

Studies demonstrate that recognizing nonverbal stimuli and cues is a predictor of higher workplace performance [18], so it is crucial to determine this skill in an organization, especially in workplaces with high mental demand in which emotion recognition and identification are done properly.

Moreover, there has not been much research on comprehending cognition instantaneously, and emotions can affect human behavior in real-time. Therefore, it seems necessary to assess the ability of CROs in emotion identification and estimation.

According to Bradley and colleagues, emotion can be divided into three main dimensions, including Valence (V), Arousal (A), and Dominance (D) [19, 20]. Valence refers to the type of emotion and defines the emotional states ranging from unpleasant (negative) to pleasant (positive) feelings. Arousal is defined as the intensity of emotion and characterizes the emotional states ranging from boredom to excitement. Dominance distinguishes emotional states with similar valence and arousal ranging from no control to complete control [11]. Besides this trilateral model, each fundamental emotion can be portrayed as a two-dimensional or bipolar model in a way that all emotions are mainly characterized by valence and arousal [11]. In addition, dominance seems to account for much less variance in semantic researches [21].

Adults typically represent multidimensional emotional traits through two primary dimensions, i.e., valence (positive-negative) and arousal (activating-deactivating) [14]. For instance, surprise evokes high arousal, and sadness has low arousal. Concerning valance, happiness and disgust represent a positive and negative valence, respectively [15]. Figure 1 illustrates a widely adopted two-dimensional arousal-valence model. The model consists of four quadrants, each representing a high or low arousal corresponding to negative or positive valence.

OPEN IN VIEWER

Fig. 1

A 2-dimensional model based on arousal and valence [14].

According to a large body of research, attention is biased toward emotional inputs when numerous stimuli compete for the organism’s resources [15]. Thus, emotional stimuli can engage attention and affect decision-making and estimating; one example is over- or under-estimating neutral situations in workplace, especially in places with high mental workloads. Therefore, measuring emotions is important, and self-reports are widely used to assess the two components of an emotional response [14]. It is imperative to know, name, understand, and label emotions appropriately. This would be more salient when considering a critical workplace such as a combined cycle power plant and realizing the importance of duties in a control room as the heart of the process.

In a nutshell, the International Affective Picture System (IAPS) is a well-known repository for emotional pictures and tagging [7]. It is a picture database where all valence and arousal levels are specified.

One of the most popular tools in the self-reported category is Self-Assessment Manikin (SAM). Because of its briefness, it may be used to record emotional reactions to a wide range of emotion elicitation methods [10, 15]. Using a self-report instrument, such as SAM, participants can manually measure their emotional responses correlated with valence from positive to negative and arousal from high to low. In fact, SAM is a tool that can be used by the test subjects to restore their emotions.

Moreover, workplaces like power plants have a crucial role in any society, not only for their products but also for the side effect of any probable danger in human health and financial domains. Given the above as well as the importance of emotion and emotion estimation, the present study aimed to investigate the impact of emotional intervention (i.e., pictures with positive and negative arousal) on the overestimation/underestimation of neutral situations based on the two-dimensional arousal-valence model. In this study, we examined how emotionally loaded stimuli might alter the subjective rating of neutral situations. It was hypothesized that emotionally loaded stimuli (as emotional intervention) would result in overestimation or underestimation of neutral situations.

3 Results

A total of 20 male subjects were screened and included in the experiment. Their ages ranged from 29 to 58 (M = 39.55, SD = 7.02), and their work experience (years) as CROs varied from 4 to 29 (14.95±6.97). Table 2 shows the participants’ demographic data.

In this research, three numerical values were considered for both arousal and valence. The first one was the image value in the IAPS database, the second one was the SAM score that a participant rated based on his subjective idea before the intervention, and the third one was the participant rating based on SAM score after intervention (Table 3). The results confirmed that there was no significant change in either arousal (p = 0.084) or valence (p = 0.189) in the SAM scores before and after the intervention.

There was no significant difference between the first and second valence (p_SAM1 &2 = 0.110) and no significant relationship between the first and second arousal (p_SAM1 &2 = 0.067).

Additionally, we found a significant difference between the mean values of IAPS arousal and SAM1 and SAM2 arousal (p_SAM1 &IAPS = 0.000 and p_SAM2 &IAPS = 0.025). The significant valence difference for SAM1 and IAPS is marginal (p_SAM1 &IAPS = 0.052). Table 4 shows all mean differences.

The study showed positive correlations at the 0.05 level amongst SAM 2 arousal ratings of participants with age (0.485) and work experience (0.461). Other demographic data had no association with SAM ratings.

This study aimed to investigate the impact of the emotional intervention on the overestimation/underestimation of neutral situations based on the two-dimensional arousal-valence model. The findings of this study showed that participants’ SAM arousal ratings (first and second) had significant differences compared to the IAPS arousal. However, SAM valence ratings had no significant differences compared to the IAPS valence, exclusive of one marginal p-value. Totally, SAM ratings in the arousal domain differed from IAPS scores in the arousal domain. This finding is similar to the result of a previous study in which the new ratings (SAM ratings) were significantly different from the IAPS score. Of course, in that study, SAM ratings in arousal level were lower than IAPS score, but in our study, SAM ratings were higher than IAPS score [22]. Similarly, a study on the effect of age on SAM ratings demonstrated that ratings were larger than the IAPS score in almost all groups [23]. Another study on the relation between virtual embodiment and cognition/emotion processing showed that participants had higher scores for arousal and valence after emotional intervention [24].

The results of comparing SAM scores revealed no significant correlation between the two SAM ratings (before and after the emotionally loaded stimuli). The results indicate that participants overestimate neutral situations only in arousal level compared with IAPS arousal. Nevertheless, after viewing emotional stimuli, participants again overestimated the neutral situations, but this estimation had fewer differences with the arousal level of IAPS. In addition, a study on SAM evaluations before and after the emotion regulation as an intervention claimed that the arousal level decreased after the emotion-based intervention. However, the valence level increased after the intervention [25]. Another study with a focus on overestimating neutral faces and related parameters revealed that participants with high social anxiety have a higher probability of misclassifying neutral emotions and valence levels (neutral emotion misclassified as anger). Moreover, this overestimation is still possible even in participants with low social anxiety. This could be a cue to assess in detail the probabilities of different ratings (such as anxiety) in the current study. Additionally, further research is required to find whether anxiety in the workplace has the chance to affect the estimation, and if it does, to what extent it modifies the estimation [26]. Previous studies have also shown some differences in SAM ratings of arousal, valence, or even both. This can be attributed to cultural differences as previously considered by other researchers [27].

According to the results of the study, CROs of the Fars combined cycle power plant could retain their ability to assess neutral situations when viewing emotionally loaded stimuli. It should be noted that this result was confirmed, especially at the valence level; however, it can be generalized based on the time interval after the exposure, which was about half an hour after the first rating. As noted in previous studies, the judgment regarding the extent of arousal and valence experienced when viewing an emotional picture using SAM is strongly associated with ratings using the semantic scales [28]. There is also evidence that participants may perform better when there is a combination of emotional stimuli, for example, auditory and visual stimuli [26].

Therefore, our hypothesis predicting that emotionally loaded stimuli (as emotional intervention) would lead to overestimating or underestimating neutral situations is confirmed in such a way that emotional stimuli can cause overestimation in neutral situations. As mentioned in the literature, valence and arousal have different effects on cognitive processes. Moreover, some researchers have emphasized the important and inevitable role of arousal in enhancing memory [29]. It is necessary to remind that understanding emotions (positive, negative, and neutral) is essential for human relationships [5]. Interaction with colleagues is very important because they spend at least one-third of the day with each other. Moreover, emotion is necessary in workplaces with human-computer interaction [24], such as in a control room, since such an interaction strongly depends on emotion and emotion recognition. Consequently, paying attention to the ability and skill of CROs in line with emotion is critical.

Nevertheless, some further issues merit comment.

First, interventions include a mixture of negative and positive images; pure negative or positive images may change the result. Second, the interval between two ratings is less than 30 minutes. As SAM-based emotion analysis suggests, less time may provide more precise results [30], and due to the importance of abrupt actions in critical workplaces, such as control rooms, emotionally loaded stimuli can quickly alter the estimation of neutral situations. This can lead to decision-making errors, improper problem-solving, and longer reaction times, mainly based on the results of studies claiming that emotion-induced subjective changes in time are common and well documented in the literature [31]. Reaction times are somewhat sensitive to emotional valence [32, 33]. This important issue indicates that in the underlying mechanisms by which emotions influence judgment and choice, significant regularities arise through various decisions [1]. Furthermore, the differences between IAPS arousal and SAM ratings in this study can be ascribed to social problems primarily related to the COVID-19 pandemic as one of the essential results to overestimate the arousal level of emotional stimuli.

Although many different studies have used questionnaires, especially SAM, to assess human emotions, this method has limitations because individuals express emotions differently [3] and objective data may solve this problem. Moreover, as a result of using SAM, the subjects may falsify or unclearly express their emotions [27].

Further research is needed to explore the interactions of human and contextual variables in emotional experience, as highlighted by researchers [8]. Consequently, it is necessary to concentrate on the contextual characteristics that may change these interactions and on how these modifications occur.

The present study has some limitations. First, all CROs were men, so the effect of gender could not be considered. Second, time was an important parameter that could affect CROs’ participation. Third, the sample size was small because of the limited number of CROs. Fourth, due to the outbreak of COVID-19, some CROs refused to participate in the study because of the fear of getting sick.

5 Conclusion

Participants overestimate neutral situations compared with IAPS only in arousal level and after emotional intervention with fewer differences from the arousal level of IAPS. Moreover, the findings suggest that CROs can still retain their ability to assess neutral situations in the case of viewing emotionally loaded stimuli, especially in valence level, at least 30 minutes after the first SAM rating. A study design with pure negative or positive high valence levels (instead of a mixture of negative and positive pictures) and a shorter duration may provide even more significant results.